Screening Topological Quantum Cathode Materials for K-Ion Batteries by Graph Neural Network and First-Principles Calculations

Abstract

Among the key parts of metal-ion batteries, cathode materials significantly affect the energy density and cycling stability. However, due to the large size of K+, not much progress has been made on cathode materials for K-ion batteries (KIBs). In this study, using the Atomistic Line Graph Neural Network and first-principles calculations, for the first time we screen cathode materials for KIBs from 7385 topological quantum materials with high electronic conductivity and reversible capacity. The experimentally synthesized K2MnS2 is discovered to have a reversible capacity of 203.8 mAh/g, an energy density of 564.5 Wh/kg, a small volume change of 6.4%, and multiple channels for K+ transport with fast dynamics. Furthermore, K2MnS2 shows high electrochemical interface stability with the reported solid electrolytes of K4V2O7, and K3NbP2O9. These findings suggest that topological quantum materials expand the design space of battery cathodes.